VNA Calibration

When I first started as an RF test engineer, I was brand new to RF measurement equipment. I had the theoretical RF framework from college and some design/simulation experience, but I never actually measured an RF circuit. On day one, I learned one of the most important aspects of measuring an RF circuit - calibrating the vector network analyzer (VNA)!

The intent of calibration is to remove the systematic measurement errors introduced between the VNA and the DUT, including cable effects, mismatch, and tracking errors. For example, say you want to measure an SMA connector on an RF device with a coax cable. You're only interested in the device itself - not the device plus the cable. A successful calibration will remove the length/loss of the coax cable, making it so the VNA only sees the DUT. When you perform a calibration, you are essentially solving for the systematic error terms of the VNA measurement system and mathematically removing them. This effectively moves the measurement reference plane from the internal VNA ports to wherever the calibration standards are defined.

The most common and widely accepted calibration methods are S.O.L.T. and TRL calibration. The S.O.L.T. (or S.O.L. for a one-port measurement) calibration includes attaching a short, an open, a load, and thru to the end of whatever coax cable is being used to measure the DUT from the network analyzer. In terms of a TRL calibration, TRL includes attaching/connecting a zero-length thru, a reflection standard (either a short or an open depending on the calibration kit), and a line standard with a known propagation delay relative to the thru. 

As with many things in life, there are tradeoffs between the two calibration methods. TRL is great for non-standard environments (such as a probe tip to a non-standard connector on a DUT), can be more precise at higher frequencies, and also has reduced reliance on calibration standards from a very expensive calibration kit. SOLT, on the on the other hand, is a quicker setup, widely supported on most VNAs, and is great for coaxial measurements when high-quality calibration kits are available.

Think of calibration like this... calibrating a VNA is not just an annoying step in the RF measurement process (even though it certainly can be); it lays the framework for accurate data acquisition.